SOCEnv / demo_task3.py
netha01's picture
Add app.py and YAML configuration
6edd399
Raw History Blame Contribute Delete
3 kB
"""demo_task3.py – Real-time visual demo of the APT Multi-Stage (Hard) task.
Streams logs step-by-step with pauses between phases to simulate
a live SOC terminal feed.
"""
import sys
import time
import io
# Force UTF-8 output on Windows
sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding="utf-8", errors="replace")
from environment import AIGymEnv
from tasks import APTMultiStageTask
from models import Action, ActionType
SEPARATOR = "-" * 62
SEVERITY_LABEL = {
"INFO": " INFO ",
"WARN": " WARN ",
"CRITICAL": " CRITICAL",
}
def print_log(log):
sev = SEVERITY_LABEL.get(log.severity, log.severity)
print(f" [{log.source.upper():<8}]{sev}: {log.message}")
def heuristic_agent(step: int):
"""Optimal SOC playbook for APT mitigation."""
if step == 1:
return Action(type=ActionType.INVESTIGATE, target_type="ip", target="198.51.100.23")
elif step == 2:
return Action(type=ActionType.BLOCK_IP, target_type="ip", target="198.51.100.23")
elif step == 3:
return Action(type=ActionType.INVESTIGATE, target_type="host", target="ws-PC042.corp")
elif step == 4:
return Action(type=ActionType.ISOLATE_HOST, target_type="host", target="ws-PC042.corp")
return Action(type=ActionType.ALLOW, target_type="ip", target="0.0.0.0")
def run_demo():
env = AIGymEnv(seed=789)
env.load_task(APTMultiStageTask())
obs = env.reset()
print()
print("=" * 62)
print(" AI SOC GYM | APT MULTI-STAGE DEMO | HARD TASK")
print("=" * 62)
print()
time.sleep(0.5)
print("[*] EPISODE STARTED — waiting for first log batch...")
time.sleep(1)
print()
print("=== INITIAL LOGS (RESET) ===")
for log in obs.logs:
print_log(log)
time.sleep(0.07)
time.sleep(1.2)
for step in range(1, 6):
act = heuristic_agent(step)
obs, reward, done, info = env.step(act)
print()
print(SEPARATOR)
print(f" STEP {step}")
print(SEPARATOR)
time.sleep(0.3)
print(f" [>] ACTION : {act.type.value.upper()} | {act.target_type} = {act.target}")
time.sleep(0.2)
print(f" [>] REWARD : {reward.score:.3f}")
time.sleep(0.2)
print(f" [>] REASON : {info.reason}")
time.sleep(0.2)
print(f" [>] EFFECT : {info.action_effect}")
time.sleep(0.8)
print()
print(" [*] Streaming new log events...")
time.sleep(0.5)
for log in obs.logs:
print_log(log)
time.sleep(0.08)
if done:
time.sleep(0.5)
print()
print("=" * 62)
print(" *** THREAT NEUTRALIZED — EPISODE FINISHED ***")
print(f" Total steps taken: {step}")
print(f" Final reward: {reward.score:.3f}")
print("=" * 62)
print()
break
time.sleep(1.2)
if __name__ == "__main__":
run_demo()